High-dimensional Data Processing And Forecasting Based On Feature Learning

The twenty-first century,with the development of computer technology,business applications and other fields has produced large amounts of data every day,in addition to a huge number of outside dimension data is also exploding.High-dimensional data in practical applications increasingly widespread,its importance is increasing,therefore,we have high dimensional data analysis and mining has a very important practical significance.Regardless of industry or academia rise to high-dimensional data analysis and mining boom.Traditional machine learning methods when faced with great difficulties to cope with high-dimensional data,a lot of low-dimensional data can superior classification algorithms in the face of high-dimensional data is difficult to meet expectations.Thus,the high-dimensional data on the traditional machine learning methods is the great challenge but also a new opportunity.How low dimensional space to show high-dimensional data,and high-dimensional data mining is an important part of the internal structure.Dimensionality reduction as an important means to overcome the "curse of dimensionality",the paper launched the feature dimension reduction in depth.There are two means of dimensionality reduction,feature selection and feature extraction.The essential difference between feature selection and feature extraction that results of feature selection is really a subset of the original feature space,and feature extraction will have new features such as a linear combination of the original features.In practice,feature selection is relatively easier to implement,but also some more applications.The main focus for the high-dimensional data research focused on the supervised classification.High-dimensional data classification have an important value.For example,the article EEG signal classification and hyperspectral remote sensing image classification and so on.For the problem is,there are still many difficulties,there are many issues we need to resolve.Therefore,the classification of high-dimensional data also launched a discussion of the characteristics of different data,it targeted choices classifier.